A New Wavelet Denoising Method for Noise Threshold
This study is proposed a new adaptive threshold based on noisy speech for each sub-bands with low complex and it is suitability for range of human hearing and range of hearing test. A new method is used wavelet 1-D experimental signal for denoising. It provided the optimal adaptive threshold of three sub-band with applies to the detail coefficients. The speech enhancement is used of threshoding on the adpated wavelet coefficients, and the results are compared a variety of noisy speech and four well-known benchmark signals. The results, measured objectively by Signal-to-Noise ratio (SNR) and Mean Square Error (MSE), are given for additive white Gaussian noise as well as two different types of noisy environment. The new method called Adaptive Thresholding with Mean for hybrid Denoising method of hard and soft function (ATMDe) and applied to hearing loss and it is found that it increases the signal-to-noise ratio by more than 114% and decreases the mean-square-error (MSE). The result of new method with SNR and MSE is higher than standard denoising methods. Hence, the new method was found that has good performance and adaptive threshold value is better than other methods.